Global Wheat Head Detection 2021: An Improved Dataset for Benchmarking Wheat Head Detection Methods

David, Etienne; Serouart, Mario; Smith, Daniel T.; Madec, Simon; Velumani, Kaaviya; Liu, Shouyang; Wang, Xu; Pinto, Francisco; Shafiee, Shahameh; Tahir, Izzat S. A.; Tsujimoto, Hisashi; Nasuda, Shuhei; Zheng, Bangyou; Kirchgeßner, Norbert; Aasen, Helge; Hund, Andreas; Sadhegi-Tehran, Pouria; Nagasawa, Kazuya; Ishikawa, Goro; Dandrifosse, Sébastien; Carlier, Alexis; Dumont, Benjamin; Mercatoris, Benoît; Evers, Byron; Kuroki, Ken; Wang, Haozhou; Ishii, Masanori; Badhon, Minhajul Arifin; Pozniak, Curtis; LeBauer, David; Lillemo, Morten; Poland, Jesse; Chapman, Scott C.; Solan, Benoît de; Baret, Frédéric; Stavness, Ian; Guo, Wei

doi:10.34133/2021/9846158

Cited by 91 publications

(63 citation statements)

References 42 publications

(24 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Here, we empirically investigate the impact of the color channel on wheat head detection and show that an appropriate treatment of color can improve detection. Specifically, given an object detector trained on the GWHD 2021 [ 7 ] dataset (e.g., we adopt Scaled-YOLOv4 [ 15 ]), we manually modify the value of each color channel using Equation ( 1 ), where α R = α G = α B = α and β R = β G = β B = β . We first fix β = 0 and vary α ( α ∈ {0.7,1.0,1.5}).…”

Section: Resultsmentioning

confidence: 99%

“…In this work, we adopt a recent Global Wheat Head Detection dataset 2021 [ 5 , 7 ] as experimental data. The RGB images in the GWHD 2021 dataset are collected between 2015 and 2020 by 16 institutions distributed across 12 countries, covering genotypes from Europe, Africa, Asia, Australia, and North America.…”

Section: Methodsmentioning

confidence: 99%

“…In addition, inconsistent labeling protocols between different datasets impede the comparison of detection methods. To tackle the issues above, the Global Wheat Head Detection dataset [ 5 , 7 ] is proposed. Based on this dataset, two sessions of the Global Wheat Challenge (GWC) have been held in the Computer Vision Problems in plant phenotyping workshops (CVPPP2020 [ 8 ] and CVPPA 2021 [ 9 ]), aimed at encouraging practitioners to develop robust algorithms.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Dynamic Color Transform Networks for Wheat Head Detection

2022

View full text Add to dashboard Cite

Wheat head detection can measure wheat traits such as head density and head characteristics. Standard wheat breeding largely relies on manual observation to detect wheat heads, yielding a tedious and inefficient procedure. The emergence of affordable camera platforms provides opportunities for deploying computer vision (CV) algorithms in wheat head detection, enabling automated measurements of wheat traits. Accurate wheat head detection, however, is challenging due to the variability of observation circumstances and the uncertainty of wheat head appearances. In this work, we propose a simple but effective idea—dynamic color transform (DCT)—for accurate wheat head detection. This idea is based on an observation that modifying the color channel of an input image can significantly alleviate false negatives and therefore improve detection results. DCT follows a linear color transform and can be easily implemented as a dynamic network. A key property of DCT is that the transform parameters are data-dependent such that illumination variations can be corrected adaptively. The DCT network can be incorporated into any existing object detectors. Experimental results on the Global Wheat Detection Dataset (GWHD) 2021 show that DCT can achieve notable improvements with negligible overhead parameters. In addition, DCT plays an important role in our solution participating in the Global Wheat Challenge (GWC) 2021, where our solution ranks the first on the initial public leaderboard, with an Average Domain Accuracy (ADA) of 0.821, and obtains the runner-up reward on the final private testing set, with an ADA of 0.695.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Dynamic Color Transform Networks for Wheat Head Detection

2022

View full text Add to dashboard Cite

show abstract

“…Another set of RGB images curated for the Sorghum Biomass Prediction Challenge on Kaggle was developed with the goal of developing methods to predict end of season biomass from images taken of different sorghum genotypes over the course of the growing season. Finally, RGB images from the TERRA-REF field scanner in Maricopa accounted for 250 of the 6000 1024x1024 pixel images in the https://www.aicrowd.com/challenges/ global-wheat-challenge-2021Global Wheat Head Dataset 2021 [11]. The goal of the Global Wheat Challenge 2021 on AIcrowd is to develop an algorithm that can identify wheat heads from a collection of images from around the world that represent diverse fields conditions, sensors, settings, varieties, and growth stages.…”

Section: Uses To Datementioning

confidence: 99%

What Does TERRA-REF's High Resolution, Multi Sensor Plant Sensing Public Domain Data Offer the Computer Vision Community?

LeBauer¹,

Burnette²,

Fahlgren³

et al. 2021

Preprint

View full text Add to dashboard Cite

A core objective of the TERRA-REF project was to generate an open-access reference dataset for the study of evaluation of sensing technology to study plants under field conditions. The TERRA-REF program deployed a suite of high resolution, cutting edge technology sensors on a gantry system with the aim of scanning 1 hectare ( 10 4 m) at around 1mm 2 spatial resolution multiple times per week. The system contains co-located sensors including a stereo-pair RGB camera, a thermal imager, a laser scanner to capture 3D structure, and two hyperspectral cameras covering wavelengths of 300-2500nm. This sensor data is provided alongside over sixty types of traditional plant measurements that can be used to train new machine learning models. Associated weather and environmental measurements, information about agronomic management and experimental design, and the genomic sequences of hundreds of plant varieties have been collected and are available alongside the sensor and plant trait (phenotype) data.Over the course of four years and ten growing seasons, the TERRA-REF system generated over 1 PB of sensor data and almost 45 million files. The subset that has been released to the public domain accounts for two seasons and about half of the total data volume. This provides an unprecedented opportunity for investigations far beyond the core biological scope of the project. This focus of this paper is to provide the Computer Vision and Machine Learning communities an overview of the available data and some potential applications of this one of a kind data.

show abstract

“…Experimental validation has been conducted on four datasets: Sim10K [22], Cityscapes [23], BDD100K [24], and GWHD [25]. We evaluate a multi-object detection scenario on a single target domain for autonomous driving, and a single-object detection scenario with multiple target domains for autonomous agriculture.…”

Section: Introductionmentioning

confidence: 99%

Domain Generalisation for Object Detection

Seemakurthy¹,

Fox²,

Aptoula³

et al. 2022

Preprint

View full text Add to dashboard Cite

Domain generalisation aims to promote the learning of domain-invariant features while suppressing domain specific features, so that a model can generalise well on previously unseen target domains. This paper studies domain generalisation in the object detection setting. We propose new terms for handling both the bounding box detector and domain belonging, and incorporate them with consistency regularisation. This allows us to learn a domain agnostic feature representation for object detection, applicable to the problem of domain generalisation. The proposed approach is evaluated using four standard object detection datasets with available domain metadata, namely GWHD, Cityscapes, BDD100K, Sim10K and exhibits consistently superior generalisation performance over baselines.

show abstract

Global Wheat Head Detection 2021: An Improved Dataset for Benchmarking Wheat Head Detection Methods

Cited by 91 publications

References 42 publications

Dynamic Color Transform Networks for Wheat Head Detection

Dynamic Color Transform Networks for Wheat Head Detection

What Does TERRA-REF's High Resolution, Multi Sensor Plant Sensing Public Domain Data Offer the Computer Vision Community?

Domain Generalisation for Object Detection

Contact Info

Product

Resources

About